InfBaGel: Human-Object-Scene Interaction Generation with Dynamic Perception and Iterative Refinement

TL;DR

InfBaGel introduces a novel framework for human-object-scene interaction generation that combines dynamic perception, iterative refinement, and hybrid training to produce realistic, consistent interactions in complex scenes.

Contribution

The paper presents a new coarse-to-fine, instruction-conditioned generation framework with dynamic perception and bump-aware guidance, addressing data scarcity and improving interaction realism.

Findings

Achieves state-of-the-art performance in HOSI and HOI generation.

Demonstrates strong generalization to unseen scenes.

Enables real-time interaction generation without detailed scene geometry.

Abstract

Human-object-scene interactions (HOSI) generation has broad applications in embodied AI, simulation, and animation. Unlike human-object interaction (HOI) and human-scene interaction (HSI), HOSI generation requires reasoning over dynamic object-scene changes, yet suffers from limited annotated data. To address these issues, we propose a coarse-to-fine instruction-conditioned interaction generation framework that is explicitly aligned with the iterative denoising process of a consistency model. In particular, we adopt a dynamic perception strategy that leverages trajectories from the preceding refinement to update scene context and condition subsequent refinement at each denoising step of consistency model, yielding consistent interactions. To further reduce physical artifacts, we introduce a bump-aware guidance that mitigates collisions and penetrations during sampling without requiring fine-grained scene geometry, enabling real-time generation. To overcome data scarcity, we design a hybrid training startegy that synthesizes pseudo-HOSI samples by injecting voxelized scene occupancy into HOI datasets and jointly trains with high-fidelity HSI data, allowing interaction learning while preserving realistic scene awareness. Extensive experiments demonstrate that our method achieves state-of-the-art performance in both HOSI and HOI generation, and strong generalization to unseen scenes. Project page: https://yudezou.github.io/InfBaGel-page/